Table 2 Unimodal and multimodal artificial intelligence-based analyses for survival prediction.

From: Multimodal artificial intelligence-based pathogenomics improves survival prediction in oral squamous cell carcinoma

  

Clinical

Pathology

Genetics

Multimodal

All features

RSF

0.714

0.530

0.529

0.722

GBSA

0.691

0.539

0.542

0.633

FastSVM

0.684

0.489

0.527

0.625

CoxPH

0.686

0.503

0.547

0.633

DeepSurv

0.462

0.538

0.501

0.515

Important features

RSF

0.698

0.635

0.637

0.834

GBSA

0.672

0.568

0.593

0.747

FastSVM

0.706

0.500

0.636

0.718

CoxPH

0.708

0.510

0.632

0.742

DeepSurv

0.413

0.557

0.503

0.635

  1. The values represent the c-index. The c-index is a commonly used metric in survival analysis that evaluates the predictive accuracy of a model. It measures the probability that, given two randomly selected patients, the patient with the worse prognosis, according to the model, will experience an event (such as death) before the patient with the better prognosis. A c-index of 0.5 indicates that the model is no better than a random chance at predicting outcomes, while a c-index of 1.0 indicates perfect predictive accuracy.
Back to article page